mail-yubao/tests/unit/split-instructions.test.ts

import { describe, expect, it } from "vitest";
import {
  detectUiKindsFromText,
  extractForecastInstructionText,
  extractMailInstructions,
  parseInstructionRoundAtMs,
  splitBodySegments,
  stripMobileMailChrome,
  extractForwardChromeMeta,
} from "@/services/parse/split-instructions";
import { matchKeywordRules } from "@/services/parse/instruction-lexicon";

const MAIL4_SUBJECT =
  "Fw: \u8f6c\u53d1\uff1a\u65b0\u589e\u9884\u62a5136\uff1a\u65b0\u8fb0\u6cfd+WHLC027G597465+WHSU8127240+\u6d1b\u6749\u77f6+40HQ+EDT2026.04-25 ETA2026.05-15\u8239\u540d\u822a\u6b21HMM EMERALD 013E+\u63d0\u62c6\u6d3e\u7ec4\u5408\u67dc-\u4e0d\u5e26\u6258\u67b6";

const MAIL4_BODY = [
  "Dear,",
  "\u539f\u7bb1\u53f7YT2604021091=FBA199R49LD6-MDW2=76\u4ef6\u64cd\u4f5c\u6307\u4ee4",
  "\u6362\u6807\u540e\u5355\u53f7\uff1aFBA19HW52S0L-2G1MCB6X-HIA1=76\u4ef6",
  "1\uff1a\u8986\u76d6\u8d34FBA\u6807\u7b7e\uff0c\u4e00\u7bb1\u8d34\u4e24\u5f20",
  "3\uff1a\u8d34\u597d\u62cd\u7167\u56de\u4f20",
].join("\n");

const MAIL4_FILES = [
  "YT2604021091=76\u4ef6\u6362\u6807\u8d70HIA1\u5361\u6d3e\u51fa\u5e93\u64cd\u4f5c\u6307\u4ee4.xlsx",
  "FBA19HW52S0L-2G1MCB6X-HIA1.pdf",
];

const MULTI_DATE_BODY = [
  "\u65b0\u589e\u8f6c\u4ed3\uff0c\u8bf7\u7559\u610f",
  "\u67dc\u53f7\uff1aMATU2745683",
  "------------------ \u539f\u59cb\u7535\u5b50\u90ae\u4ef6 ------------------",
  "\u53d1\u4ef6\u4eba\uff1aa@b.com",
  "\u4e3b\u9898\uff1a\u65b0\u589e\u9884\u62a5136\uff1a\u65b0\u8fb0\u6cfd+WHSU8127240",
  "\u8bf7\u67e5\u6536\u65b0\u589e\u9884\u62a5",
].join("\n");

describe("splitBodySegments", () => {
  it("splits on quote header", () => {
    const segs = splitBodySegments(MULTI_DATE_BODY);
    expect(segs.length).toBeGreaterThanOrEqual(2);
    expect(segs[0]).toContain("\u65b0\u589e\u8f6c\u4ed3");
  });
});

describe("matchKeywordRules", () => {
  it("maps hard words to ui kinds", () => {
    expect(matchKeywordRules("\u8bf7\u8d34\u6807\u5e76\u62cd\u7167", "body").map((h) => h.uiKind)).toEqual(
      expect.arrayContaining(["work_order"]),
    );
    expect(matchKeywordRules("\u65b0\u589e\u9884\u62a5", "subject").map((h) => h.uiKind)).toContain(
      "forecast",
    );
    expect(matchKeywordRules("\u65b0\u589e\u8f6c\u4ed3", "body").map((h) => h.uiKind)).toContain(
      "transfer",
    );
  });
});

describe("detectUiKindsFromText", () => {
  it("subject forecast", () => {
    expect(detectUiKindsFromText(MAIL4_SUBJECT, "", [])).toContain("forecast");
  });
  it("body work_order", () => {
    expect(detectUiKindsFromText("", MAIL4_BODY, [])).toContain("work_order");
  });
  it("DO filename", () => {
    expect(detectUiKindsFromText("", "", ["WHSU8127240-DO.pdf"])).toContain(
      "do_upload",
    );
  });
});

describe("do_upload dedupe", () => {
  it("does not emit two DO cards when forecast body + DO.pdf both hit", () => {
    const body = [
      "Dear,",
      "原箱号YT2604021091=76件操作指令",
      "换标后单号：FBA19HW52S0L=76件",
      "3：贴好拍照回传",
      "------------------ 原始邮件邮件 ------------------",
      "发件人：a@b.com",
      "日期：2026-05-08 19:15:53",
      "主题：新增预报136：新辰泽+WHSU8127240",
      "Dear,",
      "请查收新增预报： 新辰泽 +WHLC027G597465+WHSU8127240+ 洛杉矶",
      "+40HQ+EDT2026.04-25 ETA2026.05-15 船名航次 HMM EMERALD 013E+",
      "提拆派组合柜 - 不带托架DO也同步上传至附件请注意查收！",
    ].join("\n");
    const units = extractMailInstructions({
      subject: MAIL4_SUBJECT,
      body,
      filenames: ["WHSU8127240-DO.pdf", ...MAIL4_FILES],
    });
    const dos = units.filter((u) => u.uiKind === "do_upload");
    expect(dos).toHaveLength(1);
    expect(dos[0].isCurrent).toBe(true);
    expect(dos[0].source).toBe("body_segment");
  });
});

describe("extractMailInstructions display order", () => {
  it("sorts units newest round first by roundAt", () => {
    const body = [
      "更新派送单 地址更新 麻烦按此份为准！！！",
      "---- 转发的邮件 ----",
      "发件人江江<1817055288@qq.com>",
      "日期 2026年3月28日 下午 5:56",
      "主题 更新地址",
      "在  2026-03-20 10:00:00 ， \"op\" <op@x.com> 写道：",
      "Dear,",
      "DO请查收见附件",
      "在  2026-03-15 09:00:00 ， \"op\" <op@x.com> 写道：",
      "Dear,",
      "请查收新增预报： 新辰泽 +EGLV143661466958+TRHU8524770",
    ].join("\n");
    const units = extractMailInstructions({
      subject:
        "Fw: 新增预报136：TRHU8524770+EGLV143661466958+40HQ",
      body,
      filenames: ["TRHU8524770-DO-1.pdf"],
    });
    const kinds = units.map((u) => u.uiKind);
    const ci = kinds.indexOf("work_order");
    const doIdx = kinds.indexOf("do_upload");
    const fc = kinds.indexOf("forecast");
    expect(ci).toBeGreaterThanOrEqual(0);
    expect(doIdx).toBeGreaterThanOrEqual(0);
    expect(fc).toBeGreaterThanOrEqual(0);
    expect(ci).toBeLessThan(doIdx);
    expect(doIdx).toBeLessThan(fc);
  });

  it("mail4 work_order rounds newest first", () => {
    const body = [
      "Dear,",
      "原箱号YT2604021091=76件操作指令",
      "在  2026-05-25 13:45:03 ， \"新辰泽\" <a@b.com> 写道：",
      "Dear,",
      "WHSU8127240= 原箱号 换标",
      "在  2026-05-20 15:55:44 ， \"新辰泽\" <a@b.com> 写道：",
      "Dear,",
      "换标 167件操作指令",
    ].join("\n");
    const units = extractMailInstructions({
      subject: MAIL4_SUBJECT,
      body,
      filenames: [],
    });
    const wo = units.filter((u) => u.uiKind === "work_order");
    const times = wo.map((u) => parseInstructionRoundAtMs(u.roundAt) ?? 0);
    for (let i = 1; i < times.length; i++) {
      expect(times[i - 1]).toBeGreaterThanOrEqual(times[i]);
    }
  });
});

describe("parseInstructionRoundAtMs", () => {
  it("parses Chinese date", () => {
    expect(parseInstructionRoundAtMs("2026年3月28日 17:56")).toBeTruthy();
  });
});

describe("extractMailInstructions multi-date", () => {
  it("mail4: forecast + work_order as separate units", () => {
    const units = extractMailInstructions({
      subject: MAIL4_SUBJECT,
      body: [
        MAIL4_BODY,
        "在  2026-05-01 09:00:00 ， \"op\" <op@x.com> 写道：",
        "Dear,",
        "请查收新增预报： 新辰泽 +WHSU8127240",
      ].join("\n"),
      filenames: MAIL4_FILES,
    });
    const kinds = units.map((u) => u.uiKind);
    expect(kinds).toContain("forecast");
    expect(kinds).toContain("work_order");
    const wo = units.find((u) => u.uiKind === "work_order");
    expect(wo?.segmentText).toMatch(/操作指令|换标/);
  });

  it("keeps transfer(current) + forecast(history) as two units", () => {
    const units = extractMailInstructions({
      subject: "MATU2745683",
      body: MULTI_DATE_BODY,
      filenames: [],
    });
    expect(units.some((u) => u.uiKind === "transfer" && u.isCurrent)).toBe(true);
    expect(units.some((u) => u.uiKind === "forecast")).toBe(true);
    expect(units.length).toBeGreaterThanOrEqual(2);
  });

  it("does not duplicate subject forecast when body already has forecast", () => {
    const units = extractMailInstructions({
      subject: MAIL4_SUBJECT,
      body: "\u8bf7\u67e5\u6536\u65b0\u589e\u9884\u62a5\uff1a\u65b0\u8fb0\u6cfd+WHSU8127240+\u63d0\u62c6\u6d3e",
      filenames: [],
    });
    const forecasts = units.filter((u) => u.uiKind === "forecast");
    expect(forecasts).toHaveLength(1);
    expect(forecasts[0].source).toBe("body_segment");
  });

  it("mail4 current unit is work_order not forecast", () => {
    const units = extractMailInstructions({
      subject: MAIL4_SUBJECT,
      body: MAIL4_BODY,
      filenames: MAIL4_FILES,
    });
    const current = units.filter((u) => u.isCurrent);
    expect(current.some((u) => u.uiKind === "work_order")).toBe(true);
    expect(current.every((u) => u.uiKind !== "forecast")).toBe(true);
  });

  it("splits nested QQ inline quote headers on mail98-style thread", () => {
    const body = [
      "Forwarded message:",
      "From：朱甜甜<ztt@cnwally.com.cn>",
      "Date：2026-07-22 23:18:44(美国山区 (GMT-07:00))",
      "TCNU4412465 WL003 FBA19DYHCDLS-XPB2-37件",
      "安排卡转海，麻烦安排贴标操作。",
      "朱甜甜<ztt@cnwally.com.cn> 在 2026年7月10日 周五 17:07 写道：",
      "TCNU4412465",
      "WL103262 50433。6件",
      "这两票客户需要美时7/10 下午2点前送达",
      "朱甜甜<ztt@cnwally.com.cn> 在 2026年7月1日 周三 17:15 写道：",
      "TCNU4412465",
      "这两票总共31件记得叫仓库拍一下照片过来哈。",
      "朱甜甜<ztt@cnwally.com.cn> 在 2026年6月13日 周六 15:47 写道：",
      "TCNU4412465+ ONEYNB3BW9236300+船名航次：WAN HAI 511-V.E110+ETA:6/22",
      "拆柜清单",
      "派送要求： 1，分别按票打板 千万不要混打",
    ].join("\n");
    const segs = splitBodySegments(body);
    expect(segs.length).toBeGreaterThanOrEqual(4);
    const units = extractMailInstructions({
      subject:
        "Fw: Fw:贴标操作：派送指令：拍照指令：do请查收：TCNU4412465+ ONEYNB3BW9236300+船名航次：WAN HAI 511-V.E110+ETA:6/22+UPS-6件+亚马逊卡派-1275件+私人地址-30件+拆柜清单",
      body,
      filenames: [],
    });
    const wo = units.filter((u) => u.uiKind === "work_order");
    expect(wo.length).toBeGreaterThanOrEqual(2);
    expect(units.some((u) => u.uiKind === "work_order")).toBe(true);
    expect(units.some((u) => u.uiKind === "forecast")).toBe(true);
    expect(wo.some((u) => /贴标/.test(u.segmentText))).toBe(true);
    expect(wo.some((u) => /拍一下照片|拍照/.test(u.segmentText))).toBe(true);
    const forecast = units.find((u) => u.uiKind === "forecast");
    expect(forecast?.isCurrent).toBe(false);
    expect(forecast?.roundAt).toMatch(/2026年7月1日|2026年6月13日/);
  });

  it("adds forecast from ops subject when current body is work_order", () => {
    const subject =
      "Fw: Fw:贴标操作：派送指令：拍照指令：do请查收：TCNU4412465+ ONEYNB3BW9236300+船名航次：WAN HAI 511-V.E110+ETA:6/22+UPS-6件+亚马逊卡派-1275件+私人地址-30件+拆柜清单";
    const body = [
      "Forwarded message:",
      "From：朱甜甜<ztt@cnwally.com.cn>",
      "Date：2026-07-22 23:18:44(美国山区 (GMT-07:00))",
      "Subject：贴标操作：派送指令：拍照指令：do请查收：TCNU4412465+ ONEYNB3BW9236300+船名航次：WAN HAI 511-V.E110+ETA:6/22+UPS-6件+亚马逊卡派-1275件+私人地址-30件+拆柜清单",
      "TCNU4412465 WL003 FBA19DYHCDLS-XPB2-37件",
      "安排卡转海，麻烦安排贴标操作。",
      "注意：ups默认一件贴2张面单",
    ].join("\n");
    const units = extractMailInstructions({
      subject,
      body,
      filenames: ["WL100244-37-XPB2(1).pdf"],
    });
    expect(units.some((u) => u.uiKind === "work_order")).toBe(true);
    expect(units.some((u) => u.uiKind === "forecast")).toBe(true);
    expect(units.find((u) => u.uiKind === "forecast")?.isCurrent).toBe(false);
  });

  it("splits reply-original boundary into forecast + redirect-delivery", () => {
    const body = [
      "Forwarded message:",
      "From：km16@kmfba.com<km16@kmfba.com>",
      "Date：2026-07-21 02:44:04(美国山区 (GMT-07:00))",
      "Subject：回复：ZJYC COSU6504130784 FCIU7089474 7.10 派拆柜派送",
      " FCIU7089474 MR03002964   改地址派送到 7400 E. Slauson Ave commerce. CA 90040",
      "",
      "km16",
      "km16@kmfba.com",
      "",
      "---- 回复的原邮件 ----",
      "发送日期2026年6月26日 14:09",
      "主题ZJYC COSU6504130784 FCIU7089474 7.10 派拆柜派送",
      "",
      "FCIU7089474 提柜 COSCO 40HQ ETA：7.10 卡派拆柜派送",
      "km16",
      "",
      "km16@kmfba.com",
    ].join("\n");
    const units = extractMailInstructions({
      subject: "Fw: Fw:回复：ZJYC COSU6504130784 FCIU7089474 7.10 派拆柜派送",
      body,
      filenames: [],
    });
    expect(units.some((u) => u.uiKind === "work_order")).toBe(true);
    expect(units.some((u) => u.uiKind === "forecast")).toBe(true);
    expect(
      units.some(
        (u) =>
          /改地址派送到/.test(u.segmentText) &&
          u.uiKind === "work_order" &&
          u.keywords.includes("改地址派送"),
      ),
    ).toBe(true);
    expect(
      units.some(
        (u) =>
          u.uiKind === "forecast" &&
          /提柜 COSCO 40HQ ETA：7.10 卡派拆柜派送/.test(u.segmentText),
      ),
    ).toBe(true);
  });
});

describe("extractForecastInstructionText", () => {
  it("isolates forecast from label body", () => {
    const t = extractForecastInstructionText({
      subject: MAIL4_SUBJECT,
      body: MAIL4_BODY,
    });
    expect(t).not.toMatch(/\u8986\u76d6\u8d34/);
  });

  it("prefers modulesInstruction over foreign body segments", () => {
    const t = extractForecastInstructionText({
      subject:
        "Fw: 转发：派送要求更新: 拆柜清单更新： DO请查收+拆柜清单更新: WHSU5574991+船名航次：OOCL SINGAPORE / 065W+ETA：6/28+拆柜清单",
      body: [
        "拆柜清单更新 以此为准",
        "注意：",
        "1.WL103516 以上几票需要拦截",
        "2.WL101167 这2票改自提",
        "OOCL SINGAPORE / 065W+ETA ： 6/28+FedEx-29 件 + 拦截 -209 件 + 拆柜清单",
      ].join("\n"),
      modulesInstruction: "船名航次 OOCL SINGAPORE / 065W",
    });
    expect(t).toBe("船名航次 OOCL SINGAPORE / 065W");
    expect(t).not.toMatch(/拦截|改自提|拆柜清单更新|WL103516/);
  });

  it("does not treat ETA/船名 work-order body as forecast Instruction", () => {
    const t = extractForecastInstructionText({
      subject: "DO请查收+拆柜清单更新: WHSU5574991+船名航次：OOCL SINGAPORE / 065W",
      body: "拆柜清单更新 以此为准\n注意：需要拦截\n船名航次：OOCL SINGAPORE / 065W+ETA：6/28",
    });
    expect(t).toMatch(/船名航次.*OOCL SINGAPORE/);
    expect(t).not.toMatch(/拦截|拆柜清单更新|注意/);
  });

  it("extracts 时间/发件人/主题 from QQ forward chrome, ignores 收件人", () => {
    const body = [
      "Dear, 请处理。",
      "",
      "---- 转发的邮件 ----",
      "发件人cs3<cs3@xinfenginc.com>",
      "日期 2026年07月09日 15:32",
      "收件人fancy<fancy@link-ever.com>、receiving<receiving@usasinogroup.com>、hui",
      "<hui@usasinogroup.com>、luo<luo@usasinogroup.com>",
      "主题 TIIU8073522-90022",
      "",
      "贴标后拍照回传",
    ].join("\n");
    const meta = extractForwardChromeMeta(body);
    expect(meta.from).toMatch(/cs3.*cs3@xinfenginc\.com/);
    expect(meta.at).toMatch(/2026年07月09日\s*15:32/);
    expect(meta.subject).toBe("TIIU8073522-90022");
    // 收件人不得进 from/subject
    expect(meta.from).not.toMatch(/fancy|receiving|hui@/);
    expect(meta.subject).not.toMatch(/fancy|收件人/);

    const units = extractMailInstructions({
      subject: "Fw: TIIU8073522-90022",
      body,
    });
    const cur = units.find((u) => u.isCurrent) || units[0];
    expect(cur?.segmentSubject).toBe("TIIU8073522-90022");
    expect(cur?.roundFrom).toMatch(/cs3/);
    expect(cur?.roundAt).toMatch(/2026年07月09日/);
    expect(cur?.segmentText || "").not.toMatch(/收件人fancy|转发的邮件/);
  });

  it("strips QQ desktop 转发的邮件 header block", () => {
    const body = [
      "Dear,",
      "请贴标后拍照回传，谢谢。",
      "",
      "---- 转发的邮件 ----",
      "发件人cs3<cs3@xinfenginc.com>",
      "日期 2026年07月09日 15:32",
      "收件人fancy<fancy@link-ever.com>、receiving<receiving@usasinogroup.com>、hui",
      "<hui@usasinogroup.com>、luo<luo@usasinogroup.com>、date<date@usasi",
      "nogroup.com>、op10<op10@xinfenginc.com>、cs1<cs1@xinfenginc.com>",
      "主题 TIIU8073522-90022",
      "",
      "原业务说明可在此继续",
    ].join("\n");
    const t = stripMobileMailChrome(body);
    expect(t).not.toMatch(/转发的邮件/);
    expect(t).not.toMatch(/发件人cs3|收件人fancy|usasinogroup|主题 TIIU|日期 2026/);
    expect(t).toMatch(/请贴标后拍照回传/);
    expect(t).toMatch(/原业务说明可在此继续/);
  });

  it("strips QQ/iPhone mobile forward chrome from instruction", () => {
    const body = [
      "Fw: 转发：新增预报136：新辰泽+WHLC027G597465+WHSU8127240+洛杉矶+40HQ",
      "船名航次HMM EMERALD 013E+提拆派组合柜-不带托架",
      "【发件人75440098<75440098@qq.com>",
      "收件人3457189498<3457189498@qq.com>",
      "大小 1.4M",
      "时间 2026年7月14日 16:00",
      "发自我的iPhone】",
      "",
      "请查收新增预报资料",
    ].join("\n");
    const t = extractForecastInstructionText({
      subject: "Fw: 转发：新增预报136：新辰泽+WHSU8127240",
      body,
    });
    expect(t).not.toMatch(/发件人75440098|收件人3457189498|1\.4M|发自我的iPhone|2026年7月14日/);
    expect(t).toMatch(/新增预报|船名航次|WHSU8127240|请查收新增预报/);
  });
});